ANTMINER S7 is available at bitmaintech.com with 4.86TH/s, 0.25J/GH

[hawkfish007]

Hi Everyone.


So normally I wouldn't rant about support topics, But I have 2 major issues I have been trying to resolve over the last few days with no joy.

Now I will start by saying bitmain have normally been really good and solve issues quickly, but I don't seem to be getting anything  resolved at the moment so any advice would be much appreciated.

The first issue I have is that my latest S7 that is currently sat in customs has an invoice attached saying the value of the 1x S7 is $30,000 and they want £5000 for the VAT.

Now as far as im aware the S7 I ordered should be valued at $1650 but it seems that there has been some sort of admin error on bitmains side, I did receive an email from bitmain stating that we needed to pay our VAT or the machine would be sent back to them, I advised them of the issue and they sent over the correct invoice to me, but Fedex have request an explanation from bitmain and a copy of the invoice directly from them, so far fedex has not received a response and neither have I.

The next issue im having is with one of our current S7 machine from batch 5 has an issue,  I checked all my machines the other day and noticed a very strong electrical burning smell and noticed 2x hashing boards had gone down.

I assume they have died but cannot check as there is a warranty sticker over the machine that I do not want to touch without the permission from bitmain, So far I have emailed info@ sent a message through zendesk and also messaged zendesk when live support was available and have so far received no response.

I feel a little ignored and for the first time in 2 years im annoyed at the way Bitmain is dealing with my issues, we have easily spent well over $30,000 with them, not much compared to some bigger miners but enough to think they would as least help.

Like I said I have been more than happy with Bitmain products and there support in the past, but feel very ignored at the moment, and any advice would be greatly appreciated.

The thing that really bothers me about the S7 is the new Bitmain warranty policy stating that if anything burns on the board, there is no warranty.
It's likely that they wanted to avoid replacing boards overclocked/modded, but it also prevent honest users like you, who either got bad luck, or falling heatsinks (engineering failure) for example, to get warranty.
I hope they will reconsider it, at least for the defects caused by falling heatsinks.

If I'm not mistaken, on a batch 5, there shouldn't be any buck converters, unlike the 135 chips units.
So the most likely dead components are an asic, a capacitor, or one of the small LDOs creating the communication voltages (usually 1.8v).

You can start by shaking the unit to check if you hear a something like a heatsink that went loose.
If that's the case, you can remove one fan without losing the warranty to confirm.

You can try to negociate with them if it's a heatsink issue because there is no way they can take you as responsible for it.

Other burnt parts can be caused either by a user overclocking the unit to death, or simply by one bad component in a batch, so it's a bit harder.

If they are totally out of warranty, I'd be interested to grab one board shipped to France for experimenting.

Even though Bitmain's policy states no warranty for burned boards, Bitmain replaced 2 of my burned boards from B1 for free including shipping both ways. I didn't open the miner on my own, I emailed them that two of the boards don't hash. They asked me to open the miner and check the boards. I found loose heatsinks and burn marks. Turnaround time was less than a week after I mailed the boards.

This is a good example of customer service by Bitmain. I hope to see more experiences like this shared on the forums. While I do not like to see bad experiences we need to hear about those as well, but if you are having a reasonable response such as this, please make sure you share it! It shows improvements by bitmain, and we need to know.

Please tell us the step by step. Who you contacted and how, and who took care of your issues, sent the RMA, etc. It may help others immensely.
Thank you for sharing.

I am impressed the way Bitmain handled my warranty claim. I sent an email to info@bitmaintech describing the issues I was having, only 1 out of 3 boards worked out of the box. Wrote order number and problem on the subject line. Also added type of power supply I was using, in this case a Bitmain 1600 watter, sent a picture of input voltage with a  voltmeter just in case.  Got a reply from Sherry Liu, she asked if I opened the miner yet, if not open and inspect. I found loose heatsinks and burn marks on the boards after opening up the miner. I sent pictures to Sherry. She approved the RMA and sent me prepaid shipping labels. Note because of time difference it is not fair to expect a reply within an hour or two, wait at least 24 hours or longer depending when you sent your email. After shipping out the boards, replacement boards were at my doorstep in 7 days. I ordered more than 10 S7s from Bitmain and this is the only S7 that gave me trouble, they are all hashing at their advertised speed or faster. Also it is good to be polite, I am sure Bitmain didn't mail a defective miner, things get rough during shipping.

[icside]

Edit:
The design calculator is available, I ran some numbers assuming 200kHz switching frequency (most likely value with 2 mosfest in parallel), and 11.6 to 13.0v won't really change anything in efficiency.
And since the switching frequency is already at 200kHz, there is no gain to expect by changing it neither.

Edit2: I checked the frequency adjust resistor, and it's indeed set for 204kHz, so no gain to expect by lowering the frequency, it's already at its lowest value.

I vaguely remember reading that somebody either has seen or was hoping to see that one of RFB1 (more likely), RFB2 is settable via a micro? Anybody have a picture of the in vivo circuit?

[RichBC]

I vaguely remember reading that somebody either has seen or was hoping to see that one of RFB1 (more likely), RFB2 is settable via a micro? Anybody have a picture of the in vivo circuit?

It is discussed earlier in this thread Rfb2 to ground is adjustable as described here.

I have not fully sussed it but am making progress with the better picture. R15, 20K is not the Reference resistor it's part of the compensation network to pin 3 of the Buck with C73 & C75. The reference resistor (fb1) is R9, 82K taking the voltage from the Converter output.The resistor to ground (fb2) is R10, 6K8.

However this is where it get's interesting and I have not fully worked it out. In parallel with R10 to Ground is a series combination of R17, 12K and U2.

Which of course begs the question as to exactly what U2 (AG30 on one board AG6K on the other) My guess is that it's a single channel Digital pot controlled over I2C by the PIC, but more work needed to confirm this.

There is a  MCP4017-503E Digital Pot IC in series a 12K in parallel with the 6K8 Rfb2.

If you scan back through the thread there is further information and discussion. Another snip here.

I have done some maths on the Digital Pot, which has 32 127 positions, in combination with the 6K8 to ground & series 12K. So assuming it is a 50K Pot we get.

        Supply V   Core V
Min       8.63        0.58
Mid       9.17        0.61
Max     11.87        0.79


Which would be a very sensible range of core voltages, so may be correct?

Rich

[icside]

I vaguely remember reading that somebody either has seen or was hoping to see that one of RFB1 (more likely), RFB2 is settable via a micro? Anybody have a picture of the in vivo circuit?

It is discussed earlier in this thread Rfb2 to ground is adjustable as described here.

I have not fully sussed it but am making progress with the better picture. R15, 20K is not the Reference resistor it's part of the compensation network to pin 3 of the Buck with C73 & C75. The reference resistor (fb1) is R9, 82K taking the voltage from the Converter output.The resistor to ground (fb2) is R10, 6K8.

However this is where it get's interesting and I have not fully worked it out. In parallel with R10 to Ground is a series combination of R17, 12K and U2.

Which of course begs the question as to exactly what U2 (AG30 on one board AG6K on the other) My guess is that it's a single channel Digital pot controlled over I2C by the PIC, but more work needed to confirm this.

There is a  MCP4017-503E Digital Pot IC in series a 12K in parallel with the 6K8 Rfb2.

If you scan back through the thread there is further information and discussion. Another snip here.

I have done some maths on the Digital Pot, which has 32 127 positions, in combination with the 6K8 to ground & series 12K. So assuming it is a 50K Pot we get.

        Supply V   Core V
Min       8.63        0.58
Mid       9.17        0.61
Max     11.87        0.79


Which would be a very sensible range of core voltages, so may be correct?

Rich

[High-Fives Rich.] Thanks! Off to do my readin. I guess I might be pulling apart a unit after all one of these days.

[yslyung]

Hi Everyone.


So normally I wouldn't rant about support topics, But I have 2 major issues I have been trying to resolve over the last few days with no joy.

Now I will start by saying bitmain have normally been really good and solve issues quickly, but I don't seem to be getting anything  resolved at the moment so any advice would be much appreciated.

The first issue I have is that my latest S7 that is currently sat in customs has an invoice attached saying the value of the 1x S7 is $30,000 and they want £5000 for the VAT.

Now as far as im aware the S7 I ordered should be valued at $1650 but it seems that there has been some sort of admin error on bitmains side, I did receive an email from bitmain stating that we needed to pay our VAT or the machine would be sent back to them, I advised them of the issue and they sent over the correct invoice to me, but Fedex have request an explanation from bitmain and a copy of the invoice directly from them, so far fedex has not received a response and neither have I.

The next issue im having is with one of our current S7 machine from batch 5 has an issue,  I checked all my machines the other day and noticed a very strong electrical burning smell and noticed 2x hashing boards had gone down.

I assume they have died but cannot check as there is a warranty sticker over the machine that I do not want to touch without the permission from bitmain, So far I have emailed info@ sent a message through zendesk and also messaged zendesk when live support was available and have so far received no response.

I feel a little ignored and for the first time in 2 years im annoyed at the way Bitmain is dealing with my issues, we have easily spent well over $30,000 with them, not much compared to some bigger miners but enough to think they would as least help.

Like I said I have been more than happy with Bitmain products and there support in the past, but feel very ignored at the moment, and any advice would be greatly appreciated.

The thing that really bothers me about the S7 is the new Bitmain warranty policy stating that if anything burns on the board, there is no warranty.
It's likely that they wanted to avoid replacing boards overclocked/modded, but it also prevent honest users like you, who either got bad luck, or falling heatsinks (engineering failure) for example, to get warranty.
I hope they will reconsider it, at least for the defects caused by falling heatsinks.

If I'm not mistaken, on a batch 5, there shouldn't be any buck converters, unlike the 135 chips units.
So the most likely dead components are an asic, a capacitor, or one of the small LDOs creating the communication voltages (usually 1.8v).

You can start by shaking the unit to check if you hear a something like a heatsink that went loose.
If that's the case, you can remove one fan without losing the warranty to confirm.

You can try to negociate with them if it's a heatsink issue because there is no way they can take you as responsible for it.

Other burnt parts can be caused either by a user overclocking the unit to death, or simply by one bad component in a batch, so it's a bit harder.

If they are totally out of warranty, I'd be interested to grab one board shipped to France for experimenting.

Even though Bitmain's policy states no warranty for burned boards, Bitmain replaced 2 of my burned boards from B1 for free including shipping both ways. I didn't open the miner on my own, I emailed them that two of the boards don't hash. They asked me to open the miner and check the boards. I found loose heatsinks and burn marks. Turnaround time was less than a week after I mailed the boards.

This is a good example of customer service by Bitmain. I hope to see more experiences like this shared on the forums. While I do not like to see bad experiences we need to hear about those as well, but if you are having a reasonable response such as this, please make sure you share it! It shows improvements by bitmain, and we need to know.

Please tell us the step by step. Who you contacted and how, and who took care of your issues, sent the RMA, etc. It may help others immensely.
Thank you for sharing.

I am impressed the way Bitmain handled my warranty claim. I sent an email to info@bitmaintech describing the issues I was having, only 1 out of 3 boards worked out of the box. Wrote order number and problem on the subject line. Also added type of power supply I was using, in this case a Bitmain 1600 watter, sent a picture of input voltage with a  voltmeter just in case.  Got a reply from Sherry Liu, she asked if I opened the miner yet, if not open and inspect. I found loose heatsinks and burn marks on the boards after opening up the miner. I sent pictures to Sherry. She approved the RMA and sent me prepaid shipping labels. Note because of time difference it is not fair to expect a reply within an hour or two, wait at least 24 hours or longer depending when you sent your email. After shipping out the boards, replacement boards were at my doorstep in 7 days. I ordered more than 10 S7s from Bitmain and this is the only S7 that gave me trouble, they are all hashing at their advertised speed or faster. Also it is good to be polite, I am sure Bitmain didn't mail a defective miner, things get rough during shipping.

my case is also handled by sherry, yes i agree be patient with her, but she does reply within 24 hours or so & she even replies after office hours. once she replied me past 9pm china time.

now it's my turn to wait. package is clearing customs. let's see what does bmt engineers say. miners not oc'ed at all.

i hope it gets fixed or replaced asap.

[E]

Batch 8 on DPS-2000BB @ 743Mhz:



This is exclusive of the fans. System efficiency will depend on your cooling situation.

This works out to just over 0.27W/Gh/s - and the DPS-2000BB can do better with higher (edit for clarity:)inputline voltage

Vac       207.56
Vsupply 12.032
Vchain   10.281

I'll screenshot the miner interface once it's been running for a few hours but this is essentially 4.95TH; 0.016%HW; 60C board temps

[yslyung]

Batch 8 on DPS-2000BB @ 743Mhz:



This is exclusive of the fans. System efficiency will depend on your cooling situation.

This works out to just over 0.27W/Gh/s - and the DPS-2000BB can do better with higher input voltage

Vac       207.56
Vsupply 12.032
Vchain   10.281

I'll screenshot the miner interface once it's been running for a few hours but this is essentially 4.95TH; 0.016%HW; 60C board temps

pump up the volt on the dps, it should workout better & can go either higher or lowers the HW errors for sure.

ah i just saw it's a B8 hmmm but good to see some diff #'s.

i saw on phillipmas post & j4b's post they went higher with higher volts but of course it's the earlier batches.

i'm still happy at stock clocks but after the 90 days, he he he he !

[E]

Batch 8 on DPS-2000BB @ 743Mhz:



This is exclusive of the fans. System efficiency will depend on your cooling situation.

This works out to just over 0.27W/Gh/s - and the DPS-2000BB can do better with higher input voltage

Vac       207.56
Vsupply 12.032
Vchain   10.281

I'll screenshot the miner interface once it's been running for a few hours but this is essentially 4.95TH; 0.016%HW; 60C board temps

pump up the volt on the dps, it should workout better & can go either higher or lowers the HW errors for sure.

ah i just saw it's a B8 hmmm but good to see some diff #'s.

Yeah exactly, but I'm comfortable w/ that HW %. I don't think it reflects a condition that's likely to shorten the useful life of the unit and it's low enough not to effect utility measurably.

[J4bberwock]

Batch 8 on DPS-2000BB @ 743Mhz:



This is exclusive of the fans. System efficiency will depend on your cooling situation.

This works out to just over 0.27W/Gh/s - and the DPS-2000BB can do better with higher input voltage

Vac       207.56
Vsupply 12.032
Vchain   10.281

I'll screenshot the miner interface once it's been running for a few hours but this is essentially 4.95TH; 0.016%HW; 60C board temps

pump up the volt on the dps, it should workout better & can go either higher or lowers the HW errors for sure.

ah i just saw it's a B8 hmmm but good to see some diff #'s.

i saw on phillipmas post & j4b's post they went higher with higher volts but of course it's the earlier batches.

i'm still happy at stock clocks but after the 90 days, he he he he !

Mine is a Batch 8 too, no overvolt for now, just lucky with the chips, I guess.

[epichaberdasher]

How low can they be clocked from the firmware ?

Id like to know what the S7s consumption at the wall at its lowest setting.

[yslyung]

How low can they be clocked from the firmware ?

Id like to know what the S7s consumption at the wall at its lowest setting.

100M is available hahaha

maybe u can try & tell us the results ?

[epichaberdasher]

How low can they be clocked from the firmware ?

Id like to know what the S7s consumption at the wall at its lowest setting.

100M is available hahaha

maybe u can try & tell us the results ?

I dont have one, i would have tried it and posted the results no problem if i did

[marvykkio]

I'm waiting for me to send S7 is 3x 3x PSU
we hope that all goes well with the Italian customs, the thief that might exist

[E]

How low can they be clocked from the firmware ?

Id like to know what the S7s consumption at the wall at its lowest setting.

100M is available hahaha

maybe u can try & tell us the results ?

Sure



100MHz - 241W (2.4W/MHz)
200MHz - 422W (2.11W/MHz)
400MHz - 753W (1.88W/MHz)
700MHz - 1.30kW (1.85W/MHz)
743MHz - 1.39kW (1.87W/MHz)

But this really isn't the right PSU to supply 200-400W TomorrowSoon I'll put up DC watt numbers which will exclude PSU efficiency effects.

[yslyung]

How low can they be clocked from the firmware ?

Id like to know what the S7s consumption at the wall at its lowest setting.

100M is available hahaha

maybe u can try & tell us the results ?

Sure



100MHz - 241W (2.4W/MHz)
200MHz - 422W (2.11W/MHz)
400MHz - 753W (1.88W/MHz)
700MHz - 1.30kW (1.85W/MHz)
743MHz - 1.39kW (1.87W/MHz)

But this really isn't the right PSU to supply 200-400W TomorrowSoon I'll put up DC watt numbers which will exclude PSU efficiency effects.

Just to add in, this is on a S7 B8 on a stock DPS2000BB PSU

thx for the results. is it possible for you to up the volt on the dps to +12.5V & some data logging ?

@E do u have 162 chip to test ? & some results please if you have.

[epichaberdasher]

How low can they be clocked from the firmware ?

Id like to know what the S7s consumption at the wall at its lowest setting.

100M is available hahaha

maybe u can try & tell us the results ?

Sure



100MHz - 241W (2.4W/MHz)
200MHz - 422W (2.11W/MHz)
400MHz - 753W (1.88W/MHz)
700MHz - 1.30kW (1.85W/MHz)
743MHz - 1.39kW (1.87W/MHz)

But this really isn't the right PSU to supply 200-400W TomorrowSoon I'll put up DC watt numbers which will exclude PSU efficiency effects.

Thanks, very interesting

[E]

Just to add in, this is on a S7 B8 on a stock DPS2000BB PSU

thx for the results. is it possible for you to up the volt on the dps to +12.5V & some data logging ?

@E do u have 162 chip to test ? & some results please if you have.

Unfortunately this B8 is the only S7 I have around. Just haven't been swayed by their pricing but the B8 finally dropped to where curiosity got the better of me.

I definitely will measure at 12.5V and I'd love to get data logging setup -- just need to put hands on a gpib-usb. I've observed a few funny lower current transients with lower hashrate but no ASICs x'd out and I'd like to capture that happening in a log.

The xitron box will even transfer i.e. voltage/current waveforms sampled at 18bits/500kHz, with fsample derived from any fundamental present (allows use of rectangular windows during DSP yada yada...); just waiting for DC shunt now to be able to do direct PSU / miner system efficiency measurement (simultaneous, synchronized, high-frequency capture of V and I on AC line and on 12V line). I've had this box around for a while but I'm glad I finally got it calibrated and setup!  

[klondike_bar]

How low can they be clocked from the firmware ?

Id like to know what the S7s consumption at the wall at its lowest setting.

100M is available hahaha

maybe u can try & tell us the results ?

power consuption is mostly linear if the chip voltage is unchanged. even at 100M you might be 0.25-0.27w/GH still   (I personally know as low as 350M works on the dropdown list, no consumption numbers though)

looking at the numbers provided above, seems like at low frequencies its closer to 0.36w/GH @100Mhz (240w/670GH). server supplies likely have a 5-10% lower efficiency at <20% load, so a better psu (such as a 650W gold-ATX) might only draw 220W under the same conditions.

Consider that fans are ~5W+5W, and the controller board likely draws 5-10W extra, and the hashboard-level draw is closer to 200W, or 0.30W/GH

[nhando]

What is the normal process for Custom / Tax for these Antminer from BitMain to the US?  I'm seeing no standard methodology for Custom tax / duties, etc.  I paid $0 other than $10 to cover the receiving end for Bitmain when I went with the lower cost shipping with FEDEX.  My brother in law paid Extra to use UPS to avoid any delays with custom per Bitmain's warning on their site, but he got hit with $16 Custom charge per S7.   I thought US, we don't have to pay the custom fees per what others reported on other threads?  I'm getting ready to buy another few B9 and want to know the best way to avoid surprise charges (ie Buy via wire vs BTC, use FEDEX, don't use UPS), etc.  What's everyone's experience and what Delivery company did you use?

[wolfen]

What is the normal process for Custom / Tax for these Antminer from BitMain to the US?  I'm seeing no standard methodology for Custom tax / duties, etc.  I paid $0 other than $10 to cover the receiving end for Bitmain when I went with the lower cost shipping with FEDEX.  My brother in law paid Extra to use UPS to avoid any delays with custom per Bitmain's warning on their site, but he got hit with $16 Custom charge per S7.   I thought US, we don't have to pay the custom fees per what others reported on other threads?  I'm getting ready to buy another few B9 and want to know the best way to avoid surprise charges (ie Buy via wire vs BTC, use FEDEX, don't use UPS), etc.  What's everyone's experience and what Delivery company did you use?

I buy ALL my stuff from Yoshi in Bitmain Denver.
NEVER have a problem with customs. He imports, you get your miner shipped from Denver. For about 1 1/2 years now. Excellent customer service.